By Aaron Krol 

January 27, 2015 | The American military has struggled to provide the best healthcare to the armed forces within our complex public-private health system. The Defense Department’s modern health insurance program, TRICARE, encourages members to get their primary care in the Military Health System’s network of over 400 hospitals and medical clinics, but offers coverage at civilian health centers as well through HMO- and PPO-like plans with only modest cost shares. TRICARE plans are comprehensive and generously subsidized, and the large majority of those eligible for the program — including active duty and retired service members and their families — opt in, for a patient population of around 9.5 million.

That’s only a fraction of the patients covered by the country’s two better-known public health plans, Medicare and Medicaid. Still, TRICARE is a massive system: if it were a private insurer, it would be the second largest in the country and the largest non-profit. And unlike the Centers for Medicare and Medicaid Services, the Military Health System (MHS) administers both the TRICARE health plans and the associated primary care centers. As a result, patients in TRICARE have their medical histories stored and managed in the largest interconnected electronic health record (EHR) system in the nation, covering more patients than the California-based Kaiser Permanente hospital network or even the Department of Veterans Affairs, which has its own in-house system for VA medical centers.

The DoD’s medical data environment, the Composite Health Care System (CHCS), keeps longitudinal records on every soldier treated in military clinics, often over the course of many years. It includes modules for appointment, treatment and prescription histories, ordering drugs and lab tests, scheduling visits, ambulatory health, and even diet and nutrition. It’s also a patchwork construct put together piece by piece over decades: the oldest elements, like the semi-independent outpatient EHR, AHLTA, date as far back as the 1980s.

“Like so many systems that are custom-built, [CHCS was] built for the exigency of the day,” says Jerry Hogge, Deputy Group President for the health division of Leidos, the defense contractor that implemented and maintains the system. In the years since AHLTA was created, EHRs used in private healthcare have changed a lot, with new levels of automation, industry standards, and features that extend more visibility to patients themselves. “While we’re very proud of the support we’ve been able to provide to the DoD over the past twenty-five years… they want to take advantage of the billions that have been invested in the commercial healthcare market,” says Hogge.

Today, the DoD is preparing for a complete overhaul of CHCS, scrapping its piecemeal modules and transferring millions of patient records to a brand new system with a centralized vision and modern capabilities. That project, called DHMSM (DoD Healthcare Management Systems Modernization, pronounced dim sum), will be one of the largest federal IT contracts in history, estimated to be worth $11 billion through its fifteen-year lifespan — if it keeps within budget, a feat that few government IT projects at this scale achieve.

DHMSM also offers a rare chance to press forward on EHR reforms that the commercial space has wrestled with for years. Despite advances in EHR technology, many believe that the mass deployment of these systems has failed to deliver on promises like meaningful intercommunication between care centers, large scale data analysis to help manage health on the level of populations, and more interaction between patients and their care providers. The need to move the needle on these ambitions is especially acute in the military, which not only serves a massive population, but also has to share patients with outside care centers across the country on a daily basis.

“My hope is that there’s an acknowledgement of the opportunity to use analytics of data at a scale that is virtually impossible in any other part of the U.S. healthcare ecosystem,” says Jason Burke, an advocate for intelligent and interoperable health IT systems who currently serves as Senior Adviser for Advanced Analytics and Innovation at the University of North Carolina School of Medicine. “Huge population size and complexity, lots of geographic coverage, and a strong national imperative to do the right thing by this population — all these things come together to put some momentum around [DHMSM].”

The Defense Department is publicly committed to many of the reforms that observers like Burke are hoping to see. In opening up the DHMSM contract to bidders, the DoD has called for a system that goes beyond the major industry data standards, to provide a platform for big population and preventive health projects. Yet there are pitfalls in any IT contract of this scale, and essential features like efficient data exchange, robust security around patient information, and the thorny problem of extending an EHR to battlefield situations could end up overshadowing more forward-thinking innovations. This is further complicated by the DoD’s request for a completely off-the-shelf solution, rather than a highly customized system.

Having accepted four bids for the DHMSM project, the DoD is scheduled to award the contract this June, in time for a military-wide rollout by 2017. Its choice could have a big impact on the American healthcare system — whether leading the way on a more flexible, interconnected model of care, or further cementing the bottlenecks that hold back clinical IT today.

Communication Breakdown 

Epic, Cerner, and Allscripts are the three largest EHR vendors in the U.S. by physician usage. All three have put in bids for the DHMSM contract, in partnerships with major contractors and technology firms: Epic with IBM, Cerner with Leidos and technology services firm Accenture, and Allscripts with Hewlett-Packard and defense contractor Computer Sciences Corp. (A fourth team is also led by big contractors, but is not pitching a top commercial EHR — more on them later.)

The leading EHR companies* have all seen significant growth over the past five years, as new federal incentives have rewarded providers for going digital. This influx of money to hospitals and doctor’s offices has allowed big companies like Epic, which has roughly 20% market share across the industry, to expand further, while also making room for hundreds of new startups.

The rapid expansion of EHRs has not been uniformly welcomed. Many physicians and nurses have complained that they’re time-wasting and difficult to use, while the proliferation of vendors has caused problems when it comes time to share data and workflows across different EHR systems. A common complaint is that the leading vendors have bet big on a “lock-in” strategy where customers are kept from switching systems by the prohibitive costs of scrapping their existing records, while hospital administrators have been wooed by efficient billing systems and the ability to cash in on federal incentives, rather than capabilities that best serve patients and physicians.

That’s a perception Cerner is trying hard to change. “I think the industry has come to a pivot point,” says Travis Dalton, a VP at Cerner and General Manager of the company’s federal business projects. “There are political pressures, and external pressures around reimbursement and payment models, there are quality and cost drivers, and then there’s this conversation around interoperability… It’s putting pressure on the commercial providers, ourselves included.”

The common thread is interoperability, the power to work with outside EHRs to track a patient’s care across facilities and over a lifetime. At its simplest, interoperability demands that vendors be able to use common data formats to send basic information — a list of prescriptions, a test result, heart rate and blood pressure measurements — to care centers using competing EHRs. The industry has come together on a series of data standards for this purpose, some of which, like those developed under the umbrella of HL7 or outlined in the government’s meaningful use requirements, are now widely accepted. Yet EHR interfaces remain a bottleneck for care, with miscommunications and redundant testing regularly exposing patients to unnecessary costs and risks.

That, says Burke, is because the industry has failed to grasp the difference between data exchange and communication. “Even though we may know how to explain that somebody’s heart rate was 73, we don’t have a standard way of describing the context around which that heart rate was measured,” he says. “It would be more useful to know that your heart rate was measured as you stepped off a fifteen-minute treadmill exercise, or ten minutes after being IV infused with a particular drug.”

To provide that context, health IT systems need to routinely communicate back and forth. If a primary physician sends a patient to a specialist in another institution, both systems need to note the referral, the date a visit is scheduled, and whether the patient actually kept the appointment. If the specialist wants to prescribe a new drug, both EHRs need a chance to alert to drug conflicts or allergies. If the specialist has a question about the patient’s clinical history, she needs to be able to query the primary physician’s EHR. And the more institutions patients bounce between — say, because they’re active duty soldiers moving across the country and overseas — the more chances there are for a costly slip-up along the line.

“From an IT perspective, this is not rocket science,” says Burke. “It’s not like we need to create some entirely new technology strategy in order to get applications to share data and share workflows. Other industries have done that for a decade.” But in the healthcare industry, where even large providers may employ only a handful of IT professionals in-house, there is often less pressure from customers to have these capabilities off-the-shelf.

“A lot of the customers don’t know what they need the vendor to provide,” Burke adds. “So the first thing vendors should be doing is stepping up in more of a leadership role. They should not be waiting for the customers to tell them how to create scalable, interoperable systems.”

Beyond the Basics 

The DoD is one customer with the clout and resources to dictate terms.

“With more than half of the care provided to DoD members by providers outside of the DoD direct care system, DoD has a strong need to get information to and back from those external providers,” Kevin Dwyer, a spokesperson for the Defense Health Agency (DHA), told Clinical Informatics News. “Achieving interoperability is critical to the DoD’s requirements.”

That has bidders for the DHMSM contract anxious to show off the ways they’ve gone above and beyond adopting industry standards. At Cerner, this includes an openness to newer data standards that let users build their own workflows and programs at the intersection of EHR platforms. For instance, the FHIR standard (Fast Healthcare Interoperability Resources, pronounced fire), the latest product of HL7, breaks down clinical data into the smallest possible discrete resources (a patient, a medication, a specimen) in formats compatible with web structures like HTTP and XML. This, combined with a common application programming interface, makes it possible for users to write programs that work regardless of where data is coming from or where it’s being accessed — a big step forward from building workflows individually at each interface between health IT systems.

“We’re going beyond just two-way communications or point-to-point interfaces, and trying to look more at offering services to develop off of,” says Dalton. To that end, Cerner has been signing on to initiatives like SMART-on-FHIR and the Argonaut Project, which commit to implementing FHIR and inviting developers to build their own “apps” in FHIR-based platforms. Cerner also touts its role in partnerships like the CommonWell Health Alliance, which creates services shared between vendors to search for and identify individual patient records.

Making patient data smoothly accessible across providers is a core requirement for a widely dispersed healthcare program like TRICARE, but in a way, it’s also setting the bar very low. The DoD’s stated goal with DHMSM is to build on the foundation of interoperability to go after more ambitious health measures. One priority for the MHS is population health, using a complete view of its patients’ collective medical histories to spot early warning signs for chronic disease, and offer patients at high risk more opportunity to participate in their care. As Burke says, these kinds of efforts are only possible on a firm bedrock of interoperable records. “You cannot do population health on a silo,” he says. “You cannot manage what you cannot see.”

Dalton hopes that the emphasis on population health will be a boost for the Cerner team. “That requirement is important, and it’s well informed,” he says. “The reason you need to interoperate is both so you can deliver care at the bedside, and so you can have the breadth and scope and multitude of data to also deliver preventive care.” He stresses that Cerner, like other large EHR vendors, has increasingly invested in patient portals and wellness programs, measures that can help engage patients in preventive health.

“[The military] have enough scale, and enough breadth, and a large enough user base that they can be their own healthcare ecosystem,” Dalton adds. “So you have the opportunity, with big data and aggregated data, to create statistically significant models that allow for predictive modeling of healthcare outcomes.”

That’s been the big promise of EHRs for years, and as the field matures, major enterprise vendors like Cerner are under real pressure to make it a reality. At the same time, it’s been no small task for long-established companies that were built around managing in-house medical records to adjust to the much more strenuous demands of analytics, open data exchange, and actively informing care decisions.

Veterans Affairs 

The fourth team bidding for the DHMSM contract is a peculiar one. Led by PricewaterhouseCoopers, it includes government contractor General Dynamics IT; two EHR providers, Medsphere and DSS Inc.; and Google, which is taking a new interest in large contract work and will provide the team with both search capabilities and its expertise in cloud architecture. Google’s participation, which was announced earlier this month, is an interesting wrinkle in this bid, but the elephant in the room is the EHR the team is peddling: a custom version of VistA, the open source system that the VA built in the 1990’s.

The VA’s VistA deployment is nearly as large as CHCS, covering over 8.5 million patients. VistA is also a favorite EHR among health professionals, frequently topping surveys of user satisfaction. That’s partly because of its familiarity: the common practice of residents rotating through VA clinics means most physicians have used VistA at some point in their training. But Edmund Billings, CMO of Medsphere, believes that VistA also owes its popularity to being designed without the constraints of appealing to hospital administrators.

“[VistA] was designed to take care of our veterans, and to be very patient-centered, and the second big design goal was to be very adaptable by physicians,” says Billings. “It focused in on the key workflows to take care of the patient, and allows the doctor to do that in a streamlined fashion.” One key innovation in VistA is to maintain a single repository of data for each patient, a record that covers inpatient and outpatient visits, mental health, and even dental. Another is its suite of “procedure calls” that let users grab pieces of data from patient records without writing code for those processes.

VistA seems intuitively like a natural fit for the DoD. The VA will be the biggest point of integration with the MHS, as retired personnel transfer over to VA centers while keeping their TRICARE plans. If both departments use the same EHR technology, there should be fewer pain points transferring data back and forth or sending queries across systems. As Billings puts it, “The big question I would ask your congressman or senator is, why are we buying two systems for our warriors, when we could have one?”

Yet apparently it’s not so simple. DHMSM is actually the successor program to a failed two-year, $500 million effort to develop a shared EHR between the VA and DoD, which was abandoned in February 2013. That plan was expected to center on a “modernized” update to VistA, where service members’ records could reside from enlistment through the whole course of their lives.

In announcing the end of the integrated EHR program, then-Secretary of Defense Leon Panetta cited budgetary concerns and time constraints as the reasons the project sank. However, a Government Accountability Office report would later conclude that those claims were “not substantiated,” and that “VA and DOD lack assurance that they are pursuing the most cost-effective and timely course of action for delivering the fully interoperable electronic health record the departments have long pursued.” FedScoop, a leading government IT news source, ran a story suggesting the project collapsed amid “turf battles” between departments that had little to do with the suitability of VistA for the MHS.

Whatever the reasons DoD abandoned VistA, the PricewaterhouseCoopers team is left in the awkward position of pitching an EHR built off a system their client has already rejected. (Spokespeople for the DoD were unwilling to comment on specific bids during the procurement process, but did assure Clinical Informatics News that the process would be a “full and open competition” and all teams would be seriously considered.)

That comes on top of other disadvantages in putting together a bid of this size. The other teams have much more experience installing their systems on a huge scale: Epic, for instance, has already implemented an integrated system of over nine million patients in the Kaiser Permanente network. Medsphere’s largest client is the Indian Health Service (IHS), which serves a little under two million patients in the American Indian and Alaska Native communities — and that contract is to support an in-house version of VistA, not an implementation of Medsphere’s customized OpenVista platform from scratch.

Still, Billings believes the IHS shares a lot of the DoD’s goals for proactive healthcare, and that VistA has helped make those goals achievable. “They really run a public health model,” he says. “The foundation is population health management.” Medsphere has helped the IHS work on longitudinal tracking of patients, letting physicians see whether their patients have stuck to appointments and treatment courses, and how their visits align with health events over time. Retrieving that information, Billings says, comes down to the VistA community’s foundational commitment to interoperability.

“The definition of interoperability is not data exchange,” he says. “What we’re trying to do is open up APIs, so you can actually have continuity of care. You can share information and communication and collaboration across two systems, and that’s what you need to take better care of patients.”

Pressure on the Private System 

The PricewaterhouseCoopers team will also be trying to sell the DoD on open source architecture. The open source movement has been steadily gaining ground in the U.S. government: NASA is a big adopter, and the VA built VistA on open source code from the ground up. Nonetheless, there is some reticence in government to choose open source solutions, not least because trusted contractors with the most experience in large IT projects rarely work on open source platforms. There are also lingering security and legal concerns, especially in a sensitive area like Defense.

Medsphere is used to its model being seen as an oddity. “Healthcare is a laggard industry when it comes to open source applications,” says Billings, who describes himself as an open source true believer. “In every other industry, open source is a way of doing business at this point.” The value, he says, comes from only charging customers for support, not software licenses, and from freeing them to switch vendors if a project doesn’t pan out. “We don’t do vendor lock. Our customers have to be satisfied. Someone else who knows VistA can come in and help them, so we have to keep them happy.”

Dalton disputes that enterprise vendors like Cerner are reliant on vendor lock as a strategy, or that using closed software blocks customers from seeking out other service providers for added value. “I think that ten years ago, there was a philosophy of, the client should have our stuff everywhere, all day, and only our stuff,” says Dalton. “Over the last decade that’s changed dramatically, and our view has changed dramatically. The true value is in what you’re able to do with the data that you’re accumulating.” He adds that cost overruns can come on the support side as easily as the licensing side.

For its own part, the Defense Department appears to take the threat of vendor lock seriously, but does not see it as a reason to exclude large enterprise vendors. In a briefing shortly after the decision to choose an off-the-shelf system, Frank Kendall, Undersecretary of Defense for acquisitions, technology, and logistics, told members of the press, “[Something] we have to consider in a best-value equation is the degree to which we're locked into a specific vendor because of proprietary content and his products… We do not want to be locked into a specific vendor.”

Whatever decision the DoD makes, it will be entering a big and impactful commitment with only a partial picture of how its new EHR will pan out over years of use. Even in the hospital networks that are working hardest to make data-driven care part of their normal business today, like Kaiser Permanente and Partners HealthCare of Massachusetts, population health is still mainly in the realm of small pilot projects and speculation. For most providers, even meaningful interoperability still seems years away. While SMART-on-FHIR and CommonWell are promising initiatives, they are only just beginning to deliver any practical solutions to health practitioners.

One thing all the bidders can agree is that the consequences of DHMSM will reach beyond military health. “We view this as a great opportunity to really shape and inform healthcare on the commercial side for a long period of time to come,” says Dalton. “We’re excited about it for the industry, we’re excited about it for our company, and we feel like this fits into what we’ve been trying to do as a company for the last decade.”

Billings sees the DHMSM contract more fundamentally as a stand for open source health IT. “I think you’re seeing a movement in the government towards open source and open standards,” he says. “[DoD] could really change the whole system. If they go to closed records, and a proprietary vendor model… they’re basically going in exactly the wrong direction.”

With the chance to impact the whole industry’s priorities around interoperability, DoD needs to recognize that the stakes around DHMSM are high, and not only for the military itself. Beyond just signaling an interest in population health, the MHS needs to dive into how forward-thinking wellness programs will be designed, what interfaces with other health systems will be required, and how those needs could change over time. If the military later finds that it’s inconvenient to make certain queries or develop new cross-system workflows within the system it chose, it will be difficult if not impossible to go back.

 “These broad-based approaches to creating interoperable architectures almost always fail,” Burke warns. “You need to focus on specific use cases that allow you to enumerate what is actually needed.”

Without that attention to how its systems will be used on a daily basis, the DoD risks being just another big customer that left the details of interoperability to its vendor, and lost its chance to change the way healthcare is delivered.

Update 1/29/15: The text of this article has been edited slightly to clarify that the EHR supported by the PricewaterhouseCoopers team is not identical to the version of VistA used at the VA.

*These also include MEDITECH, GE Healthcare, CPSI, and a good half dozen others, several of which are bigger by some measures than Cerner or Allscripts. The sheer number of vendors that could be considered industry leaders is one indication of just how large this market has grown.